Stork Tracks

Route Map

Photo by Peter Schad on Unsplash

There would be no population explosion if people who are trying to keep the wolf from the door
wouldn’t let the stork fly in through the window…
— Evan Esar

Tracking data of adult white storks (Ciconia ciconia) for the years 2014–2016. The position of the storks has been estimated via GPS every 5 minutes. A stork called Gili (ID 2421/HH847) traveled the furthest distance from South Africa to Germany.

Ingest

tracking data from a directory of files

data_files <- list.files(path = "archetypes/stork-tracks/data/gps/", pattern="*.csv")
data_files

##  [1] "storks-gps-000.csv" "storks-gps-001.csv" "storks-gps-002.csv"
##  [4] "storks-gps-003.csv" "storks-gps-004.csv" "storks-gps-005.csv"
##  [7] "storks-gps-006.csv" "storks-gps-007.csv" "storks-gps-008.csv"
## [10] "storks-gps-009.csv" "storks-gps-010.csv" "storks-gps-011.csv"
## [13] "storks-gps-012.csv"

stork_gps <- purrr::map_df(data_files, 
             ~read.csv(paste0("archetypes/stork-tracks/data/gps/", .x), stringsAsFactors = FALSE) %>% 
             mutate(filename = .x))

head(stork_gps, n=10)

Wrangle

clean, parse timestamp, sort, and filter

stork_gps_wrangled <- stork_gps %>% janitor::clean_names() %>%
             mutate(timestamp = as.POSIXlt(timestamp), format = "%Y-%m-%d %H:%M:%S") %>%
             select(tag_local_identifier, individual_local_identifier, timestamp, location_long, location_lat, ground_speed, heading)


stork_gps_wrangled <- stork_gps_wrangled %>% arrange(tag_local_identifier, timestamp)

stork_gps_wrangled$year <- as.integer(stork_gps_wrangled$timestamp$year + 1900)

# Years - [1] 112 (2012) 113 (2013) 114 (2014) 115 (2015) 116 (2016)
stork_gps_year <- filter(stork_gps_wrangled, year == 2016)

head(stork_gps_year, n=10)

Wrangle

highlight Gili

gili <- filter(stork_gps_year, individual_local_identifier == '2421/HH847/Gili')
head(gili, n=10)

gili_labels <- gili %>% filter(row_number() == 1 | row_number() == n())
gili_labels

stork_gps_rest <- filter(stork_gps_year, individual_local_identifier != '2421/HH847/Gili')

Analyze

compute geospatial distance along route

# calculate the distance from previous point
gili_dist <- gili %>%
  mutate(plocation_long = lag(location_long), plocation_lat = lag(location_lat)) %>%
  rowwise() %>%
  mutate(distance = distm( c(plocation_long, plocation_lat), c(location_long, location_lat), fun=distHaversine )) %>%
  mutate(km = distance/1000)

gili_dist <- as.data.frame(gili_dist)

# calculate cumulative sum and select samples
gili_dist <- gili_dist %>% 
  filter(distance > 0) %>% 
  mutate(cumulative_km = round(cumsum(km), digits=0)) %>%
  filter(row_number() %% 750 == 1)

gili_dist

Base map, natural earth, option 1

continent and country select

africa <- ne_countries(scale = "small", continent = 'africa', returnclass = "sf")

europe <- ne_countries(scale = "small", continent = 'europe', returnclass = "sf")
europe <- filter(europe, iso_a3 != "RUS") 
europe <- filter(europe, iso_a3 != "GUF") 

middle_east <- ne_countries(scale = "small", country = c('turkey', 'syria', 'lebanon', 'jordan', 'iraq', 'israel', 'saudi arabia', 'yemen'), returnclass = "sf")

ne_region <- rbind(africa, europe, middle_east)

Base map, natural earth, option 2

world

# natural earth base map
ne_world <- ne_countries(scale = "small", returnclass = "sf")
# removes Antarctica
ne_world <- filter(ne_world, iso_a3 != "ATA")

Route map

paths, points, and labels with coordinate limits

theme_opts <- theme(
  text = element_text(family = "inconsolata"), 
  plot.title = element_text(color = "black", size = 18, face = "bold", family = "inconsolata"),
  plot.subtitle = element_text(color = "black", size = 16, family = "inconsolata"),
  plot.caption = element_text(color = "#555555", size = 8, family = "inconsolata"),    
  plot.background = element_blank(),
  panel.grid.major = element_blank(),
  panel.grid.minor = element_blank(),
  panel.background=element_rect(fill="#cacdcf", colour="#cacdcf"),
  panel.border = element_blank(),
  axis.text = element_blank(),
  axis.line = element_blank(),
  axis.ticks = element_blank(),
  axis.title = element_blank(),
  legend.position="none"
)

stork_palette <- c(
  "#F44336",
  "#E91E63",
  "#9C27B0",
  "#673AB7",
  "#3F51B5",
  "#2196F3",
  "#03A9F4",
  "#00BCD4",
  "#009688",
  "#4CAF50",
  "#8BC34A",
  "#FFEB3B"
)

v1 <- ggplot(data = ne_world) +
  geom_sf_interactive(aes(tooltip = iso_a3, data_id = iso_a3), fill="#e7e7e5", color="#717c8a", stroke=0.5) +
  geom_path( data = stork_gps_rest, aes(x=location_long, y=location_lat, group=factor(tag_local_identifier)), color = "transparent", fill = 'white', size=3, alpha = 0.6) +
  geom_path( data = stork_gps_rest, aes(x=location_long, y=location_lat, group=factor(tag_local_identifier), color = individual_local_identifier), size=0.5, alpha = 0.8) +
  geom_path( data = gili, aes(x=location_long, y=location_lat, group=factor(tag_local_identifier)), size=4, color = "transparent", fill = 'white', alpha = 0.6) +
  geom_path( data = gili, aes(x=location_long, y=location_lat, group=factor(tag_local_identifier)), size=1.5, color = 'orange') +
  geom_point( data = gili_dist, aes(x=location_long, y=location_lat), shape = 21, size = 3, color = "white", fill = "black") +
  geom_label_repel( data = gili_dist, aes(x=location_long, y=location_lat, label = paste0(cumulative_km, " kilometers\n", timestamp)), 
                    xlim  = c(max(gili_dist$location_long)+10, NA),
                    force_pull = 0, segment.size = 0.5, direction = "both", min.segment.length = 0,
                    segment.curvature = -0.1, segment.ncp = 3, segment.angle = 20,
                    size = 4, label.padding = unit(0.5, "lines"), fill = "orange", color = "black", alpha = 0.8, hjust = 0.0) +
  geom_point( data = gili_labels, aes(x=location_long, y=location_lat), shape = 21, size = 4, color = "white", fill = "orange") +
  geom_label( data = gili_labels, aes(x=location_long, y=location_lat, label = paste0("Gili\n", timestamp)), 
                    nudge_x = -6, size = 4, label.padding = unit(0.5, "lines"), fill = "black", color = "white", hjust = 0.5) +
  scale_color_manual(values = stork_palette) +
  xlim(-20,60) +
  ylim(-35,60) +
  theme_bw() +
  labs(x="",
       y="",
       title = "Stork Migration Tracks", 
       subtitle="Routes Map, 2016") +
  theme_opts
  
girafe(ggobj = v1, width_svg = 12, height_svg = 16,
       options = list(opts_sizing(rescale = TRUE, width = 0.5))
)

References

citations for narrative and data sources

Narrative: inspired by the work of Cédric Scherer
Original Research and Data source:

Rotics S, Kaatz M, Turjeman S, Zurell D, Wikelski M, Sapir N, Eggers U, Fiedler W, Jeltsch F,Nathan R (2018) Early arrival at breeding grounds: causes, costs and a trade-off with over wintering latitude. Journal of Animal Ecology.
doi:10.1111/1365-2656.12898

Rotics S, Kaatz M, Turjeman S, Zurell D, Wikelski M, Sapir N, Eggers U, Fiedler W, Jeltsch F,Nathan R (2018) Data from: Early arrival at breeding grounds: causes, costs and a trade-off with over wintering latitude.
Movebank Data Repository. doi:10.5441/001/1.v8d24552